Abstract:To address the issues of complex structure and high fabrication difficulty associated with compact multi-beam antennas in the terahertz frequency band, a terahertz Luneburg lens antenna and a partial Maxwell Fisheye(PMFE) lens antenna operating in the 355 GHz band are proposed. Both antennas use a periodic metallic bed-of-nails structure to realize a gradient refractive index surface wave lens with a minimum structural dimension of 60 μm. To regulate the beam direction in the non-scanning plane, corrugated rings are loaded on the edges of the surface wave lenses. In addition, the waveguide feeding array and the surface wave lens are designed as one piece to ensure structural integrity. High-precision 3D printing(10 μm precision) combined with magnetron sputtering are employed to realize the metallization of two surface wave lenses and low-cost realization of terahertz antenna prototype. The simulated results show the beam scanning capability of ±60° and ±72° for the Luneburg lens antenna and PMFE lens from 350 GHz to 360 GHz, respectively. The Luneburg lens antenna is verified by prototype fabrication and testing, which demonstrates good impedance matching and multibeam scanning performance from 350 GHz to 360 GHz, with a beam scanning range of ±60°, a gain higher than 16 dBi, and a beam-scanning loss better than 1.2 dB. The agreement between the measured and simulated results demonstrates the feasibility of this scheme and provides a new design idea and technology choice for the realization of terahertz multibeam antenna design.

Abstract:Facing the growing performance requirements of future mobile network application scenarios, Terahertz(THz) Integrated Sensing and Communication(ISAC) technology, with its advantages of high-speed communication and high-precision sensing, has become a current research hotspot.Existing research has largely focused on the performance analysis, waveform design, and system architecture of THz ISAC, with a lack of studies on the channel propagation characteristics of THz ISAC. To address this, a typical indoor laboratory scenario is selected, and a time-domain measurement system based on pseudo-random sequences is established to measure the indoor THz ISAC channel at 140 GHz band. Channel parameters such as path loss are analyzed based on the measurement data. The analysis results indicate that in the indoor scenario, the abundant scatterers are shared by both the communication and sensing channels. They not only appear in the results of sensing echoes but also contribute to the emergence of potential communication multipath components.

Abstract:Multiple Input Multiple Output Synthetic Aperture Radar(MIMO SAR) imaging systems use multiple channels to obtain multi-directional information about humans, which is suitable for human security scenarios. However, in THz MIMO SAR imaging systems, due to the large number of antenna elements, how to balance the accuracy and computational efficiency of the echo signal model becomes a key challenge. A cylindrical synthetic aperture is employed to irradiate the human to obtain the echo signals, and these echo signals are processed by using the Polar Format Algorithm(PFA) to achieve THz 3D human imaging. The computational efficiency and imaging results of two different echo signal models, the Physical Optics(PO) algorithm and the Ray Tracing(RT) method, are compared by simulation for computing a Perfect Electric Conductor(PEC) material-human body. The results show that the PO algorithm using Graphics Processing Unit(GPU) acceleration performs well in terms of computational efficiency and imaging quality, with less than one hour of computation time to compute the entire cylindrical synthetic aperture echo signal, and the imaging results clearly reproduce the shape of the hazardous object. In addition, the PO algorithm also performs well in calculating echo signals that match the actual material of the human body. The effect of directional antennas with different lobe widths on the imaging results is also explored. This provides a more accurate and efficient echo signal model for future validation and optimization of imaging algorithms.

Abstract:With the development of 6G networks, in-depth research on the propagation characteristics of terahertz(THz) channels in urban environments is crucial for designing efficient, reliable, and secure communication systems. The impact of different types of building corners (including acute, right, obtuse, and curved angles) on the transmission of THz channels and their physical layer security are systematically investigated through a combination of theoretical analysis, numerical simulation, and experimental measurement. The experiments are conducted by using a THz channel measurement system at three frequencies: 140 GHz, 225 GHz, and 320 GHz. Theoretical analysis is performed by using numerical simulations and knife-edge diffraction models. The research findings reveal the effects of corner structures on the propagation of THz waves, including diffraction and reflection phenomena, as well as the impact of frequency variations on these phenomena. This work provides theoretical guidance for the deployment of THz communication systems in urban environments.

Abstract:With the continuous development of 6G technology, terahertz radar and integrated sensing and communication are gradually becoming important research directions in the field of electronics and information. Programmable metasurfaces, with their advantages of being lightweight, conformable, and dynamically tunable,exhibit a high degree of freedom in terahertz beam manipulation, and thus hold significant application potential in communication, imaging, and radar. Starting from the design theory of programmable metasurfaces loaded with semiconductor components, a GaAs varactor suitable for high-frequency applications is selected, a 1 bit digital coding metasurface is constructed, and its sub-terahertz electromagnetic response and beam manipulation performance are characterized. The results show that the metasurface array has wide-angle dynamic beamforming and beam scanning capabilities in the W-band, with experimental results matching well with simulations.

Abstract:Existing millimeter-wave radar target detection is mainly implemented based on serial processing platforms, which have certain limitations in processing speed for large-sized radar images. This paper proposes a fast millimeter-wave radar target detection structure based on a Field-Programmable Gate Array(FPGA), with a chain-type First-In-First-Out(FIFO) buffer at its core and incorporating an image extraction approach. The chain-type FIFO enables multi-frame data alignment and parallel output to obtain window edge data. Based on custom window parameters, the edge data are partitioned and summed, and then delayed and cached. This allows for the reuse of computational results before and after window movement, and in combination with a pipelined processing structure, it improves computational efficiency. By reasonably partitioning the overlapping regions of adjacent sub-images and extracting multiple small-sized sub-images from large-sized images for separate processing, the implementation speed of the radar target detection algorithm is significantly increased, and on-chip logic resources are substantially conserved. The proposed FPGA-based target detection method is validated using a Frequency-Modulated Continuous Wave(FMCW) millimeter-wave radar operating in the 92~ 94 GHz band. For a large-sized radar image of 1,000×2,000 pixels, a rapid processing time of 120 ms is achieved. The deployed FPGA algorithm consumes only thirty-two 18K BRAMs and 6 461 LUTs.

Abstract:Unmanned Aerial vehicle ad-hoc Network(UANET) can increase the communication range by multi-hop forwarding, in which the routing algorithm undertakes the task of packet transmission path planning. To address the gain attenuation problem caused by inaccurate directional antenna beam pairing due to UAV positioning deviation in highly dynamic networks, an Ant Colony Optimization routing algorithm based on Link Quality Prediction(LQP-ACO) is proposed. The algorithm first predicts the link quality between UAV nodes using Bidirectional Gated Recurrent Unit-Fully Connected Neural Network(BiGRU-FCNN). Then, based on the predicted link quality, ant colony optimization algorithm is employed to find the two optimal paths for business data transmission. Simulation results show that the routing algorithm proposed in this paper reduces the packet loss rate by 2.75% and 4.5% respectively compared to the traditional Dijkstra's algorithm under Random Way Point(RWP) as well as Random Walk(RW) mobile models.

Abstract:The identification of radiation source threat level is an important basis for interference resource allocation. Currently, the commonly used threat level judgment methods in engineering only consider a single method for indicator weighting, which has strong subjective factors and often results in unreasonable results. In response to the above issues, this article proposes a subjective and objective comprehensive weight threat level identification method based on game theory. This method utilizes four threat indicators, namely carrier frequency, pulse width, pulse repetition period and duty cycle, to establish corresponding membership functions. Using the ideas of game theory, the weights obtained from tomography analysis and entropy method are combined to obtain comprehensive weights; combining weight with membership degree, the threat level recognition results are obtained. This article analyzes the parameters of radar radiation sources in multi-environment battlefields, and the results show that using a threat level evaluation method based on game theory can provide more reasonable results compared to using only a single evaluation method, achieving good recognition results for multi-radiation source scenes.

Abstract:Aiming at the synchronization errors between distributed positioning sensors equipped with single antennas, a multi-frequency fusion Direct Positioning Determination(DPD) algorithm based on a single auxiliary source is proposed. Firstly, the received signals are segmented, and the frequency-domain data of the wideband signals are converted into a set of single-frequency signals. Secondly, the synchronization error matrix is pre-compensated using the signals received from the single auxiliary source, and this matrix is then applied to the received data from the target radiation source. Next, the subspace orthogonality relationships under multiple single-frequency signals are fused, and a minimized cost function that separates the unknown attenuation coefficients is constructed using the Lagrange multiplier method. Finally, the target area is divided into a grid, and the radiation source is estimated through peak searching of the cost function. Simulation results validate the effectiveness of the proposed algorithm. Under given parameter conditions, compared with another single-auxiliary-source-based direct positioning algorithm, the computational complexity of this algorithm is reduced by at least 86.91%, achieving better positioning accuracy under different Signal-to-Noise Ratio(SNR) and synchronization error conditions.

Abstract:The Coupling Cross Section(CCS) of aperture is an important parameter to evaluate the effect of aperture penetration. Using BP neural network to predict CCS has a much higher prediction speed than full-wave analysis and better accuracy than traditional formula methods. This paper focuses on the prediction model which can be applied to multi-shape aperture array. Three neural network models are proposed to predict the CCS of aperture array, including one traditional single-stage model and two two-stage models. Taking the regular hexagonal aperture array as an example, the performance of the three models is compared. These results show that the double-level model with the most prior information performs the best. The Root Mean Square Error(RMSE) of the CCS prediction for the regular hexagonal aperture array by this model is 0.017 2, and the coefficient of determination(R) is 0.999 1. When this model is transferred, it can predict the CCS of circular and square aperture arrays, with an average relative error of 1.94% for the samples. The prediction results confirm the precision, efficiency, and universality of the model.

Abstract:To increase the speed of calculating the Radar Cross Section(RCS) in the optimization design process for reducing the RCS of electrically large targets, a multi-layer Fully Connected Neural Network(FCNN) is trained using the results of models calculated by electromagnetic simulation software when employing heuristic algorithms for low-RCS optimization design of electrically large targets. During the optimization process, once the number of calculated models is sufficient to complete the training of the neural network, the trained neural network is employed to improve electromagnetic simulation calculations. Leveraging the faster computational speed of neural networks compared to electromagnetic simulations, the optimization design speed for low-RCS of electrically large targets is enhanced. Under the conditions of the electrically large target model selected in this paper and the optimization design using the simulated annealing method, the use of a multi-layer fully connected neural network to improve electromagnetic simulation calculations significantly increases the speed of low-RCS optimization design, reducing the required time from over 300 h to approximately 140 h.

Abstract:The current extraction of image information is constrained by the transmission of massive data and the limitations of channel communication capabilities. To address this, a multi-layer image information extraction system has been constructed to overcome the limitations of transmission time and communication capacity. Based on information entropy theory, a multi-layer target image information extraction algorithm is established, using the minimum information entropy of image-based sensor information as input. By combining image feature engineering, the algorithm enhances the feature extraction and inference of images. It also utilizes sensor image data to extract the minimum volume of key information from images. Experiments have verified that this algorithm can increase the image information compression ratio to 10⁶ without losing complete and effective information. This effectively solves the problem of real-time and reliable long-distance image information transmission with low data volume and low bandwidth.

Abstract:To avoid the phenomenon of significant load differences in dedicated transformer acquisition terminals affecting their operational stability, an Adaptive Load Balancing(ALB) algorithm for dedicated transformer acquisition terminals based on a dynamic coordinate transformation algorithm is proposed. The load information of the dedicated transformer acquisition terminals is collected using a distributed information acquisition method, and the collected load information is quantitatively processed through a centralized migration method. By combining the weighted average algorithm with the bilinear interpolation of the residual of the dedicated transformer acquisition terminal nodes, new coordinates for the dedicated transformer acquisition terminals are obtained through dynamic coordinate transformation. Based on the coordinate positions of the dedicated transformer acquisition terminals, a star structure is established for each terminal node. Combined with the collected load information, the relationship between the load and the load transfer threshold is employed to determine whether the load of the dedicated transformer acquisition terminal nodes is balanced. When the load is unbalanced, it is necessary to identify overloaded nodes and establish a backup node table based on a binary tree to transfer the load from the overloaded nodes to the backup nodes, thereby achieving adaptive load balancing of the dedicated transformer acquisition terminals. Experimental results show that the application of this algorithm can effectively alleviate the pressure on overloaded nodes and achieve a balanced load state for the dedicated transformer acquisition terminals. The coordinate transformation error is relatively small. It can effectively reduce the response time for load balancing of the dedicated transformer acquisition terminals, enhance throughput, and improve load balancing performance, resulting in a good load balancing effect.

Abstract:The problem of large power loss rate exists both in the design of distribution network route structure and in the layout of outlets. In order to reduce the power loss of the distribution network and continuously improve its power quality, the active power adjustment method of distribution network group-controlled station area based on Automatic Voltage Control(AVC) technology is studied. Based on the equivalent load of distribution network line operation, the unbalanced power component of distribution network group-controlled station area is defined; the negative sequence component and positive sequence component are divided according to the dynamic command corresponding to the operation mode, and the power regulation function of distribution network group-controlled station area is determined under different differential gain conditions; according to the regulation function and the ranges of the different parameters, the regulation mode is determined based on the AVC technology, and the power quality of distribution network group-controlled station area is improved by AVC technology. According to the regulation function and the range of different parameters, the regulation mode is determined based on AVC technology to actively regulate the unbalanced power of distribution network group-controlled stations, and the method design is completed. Taking several public stations in the distribution network as the test object, and by adjusting the reactive power compensation capacity of each group of stations, the compensation capacity can reach up to 580 kVA. It indicates that the proposed method can realize the active regulation of the power capacity, meeting the operational standards of distribution network lines and demonstrating practical application value.

Abstract:When college students are under excessive psychological stress, they may experience severe psychological problems, necessitating effective methods to predict their mental health development. Based on this, the prediction algorithm is studied for the deviation trend of college students' mental health development. By quantifying college students' psychological data through weighted association rules and characterizing the development trend of psychological data using a multi-edge neighborhood matrix, the prediction of the deviation in the mental health development trend of college students is achieved based on the autocorrelation function, completing the design of the prediction method. Experimental results show that, using different stress events as test content affecting the mental health development of college students and comparing predictions through mean absolute error, mean squared error, and root mean squared error, the new method can achieve trend prediction with minimal error, with error values all below 0.1. This method can ensure an accurate judgment of the mental health development of college students and has practical application value.